Resource transfer control method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present application disclose a resource transfer control method and device, electronic equipment and storage medium. The method comprises: determining a target operation intention for a resource transfer service and at least one user demand constraint combination according to service dialogue information; indexing a corresponding resource transfer service processing workflow from a preset job graph based on the target operation intention, the resource transfer service processing workflow comprising a plurality of subtask nodes arranged according to a preset logical topology and executed by at least one functionally heterogeneous large model agent; driving the at least one large model agent to sequentially execute each subtask node to generate at least one resource transfer control scheme based on the user demand constraint combination and a resource transfer selection item library; and generating and outputting reply information based on a dialogue interaction agent according to the resource transfer control scheme. The method can greatly simplify the user's operation process in the resource transfer process and improve the operation flexibility.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a resource transfer control method, apparatus, electronic device, and storage medium. Background Technology

[0002] In resource transfer scenarios, the transferred resource data can be any schedulable, quantifiable, cost-based, and quota-constrained transferable data, including but not limited to computing power data, electricity data, and carbon energy data. Users can leverage a service platform that provides resource transfer services (i.e., the service platform is associated with a "platform data account") to transfer resource data from one data source node (i.e., the data source node is associated with an "organizational data account") to another data source node.

[0003] However, currently, when users transfer resource data through relevant service platforms, the operation process is often cumbersome. For example, users need to fill out forms or enter multiple links to obtain a suitable solution for resource transfer, resulting in low operational flexibility. Summary of the Invention

[0004] This application provides a resource transfer control method, apparatus, electronic device, and storage medium, which can greatly simplify the user's operation process and improve operational flexibility during resource transfer.

[0005] In a first aspect, embodiments of this application provide a resource transfer control method, including: In the resource transfer scenario, in response to the service dialogue information input by the user terminal to the dialogue interaction agent, the target operation intention for the resource transfer service and at least one combination of user demand constraints are determined based on the service dialogue information. Based on the target operation intent, the corresponding resource transfer service processing workflow is indexed from the preset operation graph. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions. Drive at least one large model agent to sequentially execute each of the sub-task nodes in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library; Based on the dialogue interaction agent, resource transfer control response information is generated according to the resource transfer control scheme, and the resource transfer control response information is output to the user terminal.

[0006] Secondly, embodiments of this application provide a resource transfer control device, comprising: The determining unit is used, in a resource transfer scenario, to respond to service dialogue information input by the user terminal to the dialogue interaction agent, and to determine the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information. An indexing unit is used to index the corresponding resource transfer service processing workflow from a preset job graph based on the target operation intention. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions. A driving unit is used to drive at least one large model agent to sequentially execute each of the sub-task nodes in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library. The response unit is used to generate resource transfer control response information based on the dialogue interaction agent and the resource transfer control scheme, and to output the resource transfer control response information to the user terminal.

[0007] Thirdly, embodiments of this application provide an electronic device, including: a processor and a memory; The processor is connected to the memory. The aforementioned memory is used to store executable program code; The processor reads the executable program code stored in the memory to run the program corresponding to the executable program code, so as to execute the method provided by the first aspect of the embodiments of this application or any possible implementation of the first aspect.

[0008] Fourthly, embodiments of this application provide a computer storage medium storing a plurality of instructions, which are adapted to be loaded by a processor and executed by the method provided in the first aspect of the embodiments of this application or any possible implementation thereof.

[0009] Fifthly, embodiments of this application provide a computer program product containing instructions that, when run on a computer or processor, cause the computer or processor to execute the method provided by the first aspect of the embodiments of this application or any possible implementation thereof.

[0010] The beneficial effects of the technical solutions provided in some embodiments of this application include at least the following: In one or more embodiments of this application, in response to service dialogue information input by the user terminal to the dialogue interaction agent, the target operation intention for the resource transfer service and at least one combination of user demand constraints are determined based on the service dialogue information; then, based on the target operation intention, the corresponding resource transfer service processing workflow is indexed from a preset job graph. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model agent with heterogeneous functions; subsequently, at least one large model agent is driven to sequentially execute each sub-task node in the resource transfer service processing workflow to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library; finally, based on the dialogue interaction agent, resource transfer control response information is generated according to the resource transfer control scheme and output to the user terminal. In other words, on the one hand, by completing the closed loop of demand collection and result feedback through conversational intelligent agents, the resource transfer service has been transformed from being driven by manual operation to being driven by intelligent dialogue. Users can obtain resource transfer control solutions that meet their needs simply through natural dialogue, without having to manually fill out cumbersome forms or navigate through multiple links. This greatly simplifies the user's workflow during resource transfer, reduces user interaction costs, and improves the flexibility and user experience of resource transfer operations. On the other hand, by standardizing the workflow of resource transfer services through a job graph, and combining it with functionally heterogeneous large-scale intelligent agents to complete the specialized division of labor for sub-task nodes, this collaborative model not only achieves modular and standardized decomposition of the resource transfer service process, ensuring the standardization of process execution, but also effectively guarantees the processing accuracy and efficiency of each sub-task by relying on the heterogeneous capabilities of large-scale intelligent agents. This makes the execution logic of the resource transfer service more flexible and adaptable to different scenarios, providing reliable technical support for resource transfer services under different demand scenarios. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic diagram of the system architecture of a resource transfer control method provided in an exemplary embodiment of this application; Figure 2 A flowchart illustrating a resource transfer control method provided for an exemplary embodiment of this application; Figure 3 A schematic diagram of the overall framework structure of a resource transfer control scheme generation process provided for an exemplary embodiment of this application; Figure 4 A flowchart illustrating another resource transfer control method provided for an exemplary embodiment of this application; Figure 5 A flowchart illustrating another resource transfer control method provided for an exemplary embodiment of this application; Figure 6 A schematic diagram of the structure of a resource transfer control device provided for an exemplary embodiment of this application; Figure 7 A schematic diagram of the structure of an electronic device provided for an exemplary embodiment of this application. Detailed Implementation

[0013] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0014] The terms "first," "second," "third," etc., used in this specification, claims, and the foregoing drawings are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or apparatus.

[0015] This application provides a resource transfer control method, apparatus, electronic device, storage medium, and computer program product.

[0016] Please see Figure 1 , Figure 1 This is a schematic diagram of the system architecture of a resource transfer control method provided for an exemplary embodiment of this application. For example... Figure 1 As shown, the system architecture may include a terminal 101, a network 102, and a server 103. The network 102 serves as the medium for providing a communication link between the terminal 101 and the server 103. The network 102 may include various types of wired or wireless communication links, such as wired communication links including fiber optic cables, twisted-pair cables, or coaxial cables, and wireless communication links including Bluetooth communication links, Wireless-Fidelity (Wi-Fi) communication links, or microwave communication links, etc.

[0017] Terminal 101 is a user-end electronic device, including but not limited to laptops, mobile phones, tablets, laptops, and desktop computers. Server 103 can be a backend server of a service platform that can provide resource transfer services for resource data, including but not limited to computing power data, electricity data, and carbon energy data. Terminal 101 can interact with server 103 via network 102 to receive or send messages to server 103, or terminal 101 can interact with server 103 via network 102 to receive messages or data sent to server 103 by other users.

[0018] In this embodiment, a dialogue-based intelligent agent for user interaction can be deployed within the service platform. Users can interact with the dialogue-based intelligent agent through input devices mounted on terminal 101 in a dialogue format. Dialogue methods include, but are not limited to, multimodal formats such as voice and text. Input devices include keyboards, microphones, and touchscreens. In resource transfer scenarios, the core task of the service platform is: In response to service dialogue information input by the user to the dialogue interaction agent, the system determines the target operation intent for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information. Each user demand constraint combination includes constraints corresponding to the user's core needs and at least one other type of constraint, including but not limited to resource transfer amount (e.g., 1000 kilowatts of electricity), resource usage scenarios, and expected user benefits (e.g., computing power deduction benefits, data transfer cost deduction benefits, data transfer points redemption benefits). Next, based on the target operation intent, the system indexes the corresponding resource transfer service processing workflow from a pre-defined job graph. This workflow includes multiple sub-task nodes arranged according to a pre-defined logical topology and executed by at least one large-scale model agent with heterogeneous functions. Then, the system drives at least one large-scale model agent to sequentially execute each sub-task node in the resource transfer service processing workflow to generate at least one resource transfer control scheme based on the user demand constraint combination and the resource transfer option library. Finally, based on the dialogue interaction agent, resource transfer control response information is generated according to the resource transfer control scheme, and the resource transfer control response information is output to the user terminal (such as displaying text / image response information on the display screen on the user terminal, or playing voice response information through the speaker, etc.).

[0019] It should be understood that Figure 1 The number of terminals 101, networks 102, and servers 103 in the diagram is only illustrative. Depending on the implementation requirements, there can be any number of terminals 101, networks 102, and servers 103.

[0020] Based on the above system architecture, this application provides a resource transfer control method. Please refer to... Figure 2 and Figure 3 , Figure 2 A flowchart illustrating a resource transfer control method provided as an exemplary embodiment of this application is shown below. Figure 3 This is a schematic diagram of the overall framework structure of a resource transfer control scheme generation process provided for an exemplary embodiment of this application. The executing entity of this resource transfer control method can be a service platform, which can provide resource transfer services for resource data, including but not limited to computing power data, electricity data, carbon energy data, etc.

[0021] Specifically, the resource transfer control method includes at least the following steps S202-S208, wherein: S202. In the resource transfer scenario, in response to the service dialogue information input by the user terminal to the dialogue interaction agent, determine the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information.

[0022] The dialogue interaction agent is a user-interaction-oriented intelligent agent deployed in the service platform, possessing core functions such as dialogue guidance, semantic understanding, and response generation. Service dialogue information can be raw input information (such as raw text input) collected during user interaction with the dialogue interaction agent, or pre-processed input information. The dialogue interaction method can be selected according to needs, such as voice interaction, text interaction, or image-text interaction. Pre-processing operations include, but are not limited to, speech-to-text conversion, image-text content extraction, standardization of multi-format input content, and filtering of invalid and redundant information.

[0023] The target operation intent is the core user operation intent obtained by intent recognition from service dialogue information, including but not limited to resource transfer intent, strategy optimization intent for resource transfer control scheme, and execution confirmation intent for transfer instructions. Resource transfer intent refers to the need to generate a resource transfer control scheme for resource transfer; execution confirmation intent refers to the need to verify and confirm the execution of the resource transfer control scheme; and strategy optimization intent refers to the need to adjust parameters (such as resource transfer amount, resource transfer cost, etc.) of the existing resource transfer control scheme (essentially also a scheme generation process).

[0024] Optionally, when the target operation intent is strategy optimization, the user requirement constraint combination can be only one, and only includes the constraint conditions corresponding to the user's core demands (such as the parameter items that need to be optimized); when the target operation intent is resource transfer, the user requirement constraint combination can be multiple, and includes the constraint conditions corresponding to the user's core demands, as well as the constraint conditions of at least one other demand type. The demand types covered include, but are not limited to, resource transfer amount (the amount of resource data to be transferred) demand, resource transfer cost demand, user rights demand, user preference demand, and resource return demand.

[0025] Specifically, resource transfer quota requirements are used to limit the amount of resource data to be transferred, such as 100 TFLOPS of computing power or 1000 kilowatts of electricity; user rights requirements are used to limit the additional rights brought about by resource transfer behavior, including but not limited to computing power deduction rights, data transfer cost deduction rights, data transfer points redemption rights, electricity price coupons, etc.; resource transfer cost requirements are used to limit the various data transfer costs, data transfer expenses and related data transfer costs to be paid for resource transfer behavior; user preference requirements are personalized requirements based on users' personal usage habits, such as preferred online operation channels for resource transfer or preferred specific service providers for computing power scheduling; resource return requirements are used to limit the relevant requirements involved in the resource payment-return closed loop flow behavior, including but not limited to resource return cycle and resource return method (such as computing power returned in a tiered manner on a quarterly basis, and electricity returned in a one-time manner on an annual basis).

[0026] For example, please continue to see Figure 3 If, during a dialogue with the interactive intelligent agent, the user's intended operation is identified as a resource transfer intention, and the specific demand type is a resource transfer quota demand, then if each user demand constraint combination contains only two types of constraints, several user demand constraint combinations can be formed, namely: resource transfer quota demand + user rights demand, resource transfer quota demand + user preference demand, resource transfer quota demand + resource return demand, and resource transfer quota demand + resource transfer cost demand. Each user demand constraint combination will subsequently output a corresponding resource transfer control scheme, and this resource transfer control scheme must meet the most basic resource transfer quota demand, thereby achieving the goal of recommending the most suitable solution to the user from multiple dimensions based on the user's core needs.

[0027] It should be noted that when deploying the dialogue interaction agent, a related knowledge base and toolkit deeply adapted to the resource transfer scenario must be configured for the agent to call. This related knowledge base covers knowledge in areas such as resource transfer operation rules, risk control strategies, historical interaction cases, and compliance requirements, providing core basis for the agent's semantic understanding, intent recognition, and decision output. The toolkit may include data transfer support tools and data transfer scheme generation tools. The support tools are used to execute resource transfer according to the generated resource transfer control scheme after user qualification verification is passed. The scheme generation tool is used to generate a resource transfer control scheme that conforms to a standard format.

[0028] S204. Based on the target operation intent, index the corresponding resource transfer service processing workflow from the preset operation map. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions.

[0029] The task graph includes processing workflows corresponding to various types of operation intentions, and each processing workflow is associated with a unique intention label. As a structured storage and indexing carrier for workflows, the task graph uses a graph structure to organize various processing workflows. The nodes in the graph include intention nodes, workflow nodes, subtask nodes, and agent nodes, while edges are used to represent the relationships between nodes (such as the mapping relationship between intention nodes and workflow nodes, the inclusion relationship between workflow nodes and subtask nodes, and the execution relationship between subtask nodes and agent nodes).

[0030] The specific process of matching target workflows (such as the resource transfer service processing workflow mentioned above) in the job graph through indexing is as follows: First, the target operation intent is standardized (e.g., redundant semantics are removed and intent expression is standardized) to generate intent keywords that conform to the job graph indexing rules; then, based on the intent keywords, the intent nodes in the job graph are traversed to locate the intent node that matches it precisely; finally, the corresponding processing workflow is indexed through the edges associated with the intent node to complete the rapid retrieval of the workflow.

[0031] It should be noted that the multiple sub-task nodes in the resource transfer service processing workflow in the job graph are arranged according to a preset logical topology (such as serial execution, parallel execution, branch judgment execution, loop execution, etc.), and each sub-task node is executed by a large model intelligent agent with heterogeneous functions. That is, the intelligent agents corresponding to different sub-task nodes have differentiated core functions (such as the following library screening intelligent agent, solution interpretation intelligent agent, parameter processing intelligent agent, etc.). Each intelligent agent ensures the orderly progress of the entire resource transfer service process by cooperating in the execution of sub-tasks.

[0032] S206. Drive at least one large model agent to sequentially execute each sub-task node in the resource transfer service processing workflow, and generate at least one resource transfer control scheme based on the combination of user requirement constraints and the resource transfer option library.

[0033] The resource transfer option library is a scenario-based product library built by the service platform for resource transfer scenarios. Each product in the library is configured with basic attribute information, including but not limited to the product's core parameters, such as the maximum resource transfer amount supported by the product, the resource transfer cost to be charged, the allowed resource transfer time period (e.g., immediate or next-day transfer), the resource return cycle, the resource return method, and the user benefits provided by the product (e.g., points vouchers); resource transfer rules, such as the qualification requirements for eligible users (e.g., enterprise electricity consumption level and power supply qualification in the power sector, enterprise computing power demand level and cloud service qualification in the computing power sector), the operational procedures for resource transfer, and the daily / monthly transfer limit. The basic attribute information may also include the resource transfer channel, default and supporting rules (e.g., overdue penalty rules, penalty standards for early termination of resource transfer, and handling mechanisms for abnormal resource transfers). Optionally, the resource transfer option library supports dynamic updates and can optimize the content within the library based on scenario iterations, policy updates, and changes in user needs. The configuration management intelligence in the platform's backend is responsible for the entire lifecycle management, ensuring the timeliness, accuracy, and availability of product and attribute information within the library through regular verification, adding product configurations, and eliminating invalid configuration items.

[0034] Please continue reading Figure 3The execution process of the resource transfer service processing workflow mainly includes the following core steps: First, filtering the selection options in the library, i.e., driving the library filtering agent, performs multi-dimensional matching and filtering of the selection options in the resource transfer option library based on user requirement constraint combinations, eliminating resource transfer selection options that conflict with constraint conditions or do not meet operational compliance standards, thus obtaining the target resource transfer selection options corresponding to each user requirement constraint combination; Second, constructing the solution generation input parameters, i.e. driving the parameter processing agent, based on the target resource transfer selection options, completes the organization, verification, and construction of the solution generation input parameters, ensuring that the input parameter information is complete, formatted correctly, and meets the requirements of subsequent solution generation. The process involves four steps: first, invoking the tool; second, generating the solution based on the input parameters of the solution generated in the previous step; third, generating the solution using the built-in algorithm model and rule engine; and fourth, interpreting and verifying the solution, which involves driving the solution interpretation agent to provide a professional interpretation of the resource transfer control solution, summarizing the core terms of the solution (such as resource transfer amount, resource return cycle, resource transfer cost, etc.) and converting them into user-friendly expressions. At the same time, it links with the security verification module to comprehensively verify the compliance and security of the solution, identify potential risks, and ensure that the solution meets regulatory requirements and user resource security standards.

[0035] It should be noted that during the execution of multiple sub-task nodes, the large model agents with different functions can communicate with each other through the platform's agent collaboration and interaction mechanism to ensure the continuity and consistency of sub-task execution. If problems such as unclear constraints are found during the execution of a certain sub-task node, the feedback will be sent to the dialogue interaction agent, which will then initiate supplementary inquiries to the user to obtain key information before continuing to advance the sub-task execution.

[0036] S208. Based on the dialogue interaction intelligent agent, generate resource transfer control response information according to the resource transfer control scheme, and output the resource transfer control response information to the user terminal.

[0037] The dialogue interaction agent can integrate the solution interpretation results with the resource transfer control solution that has passed security verification. According to the preset reply template (adapted to different types of operation intentions) and the script specifications, it generates resource transfer control reply information and embeds necessary interactive guidance (such as confirmation of execution instructions, supplementary material prompts, abnormal situation explanations, etc.) to ensure that the final reply content is clear, accurate and complete, enabling users to quickly understand the core information and adaptability of each solution.

[0038] Optionally, when outputting response information to the user, the output format and display style can be adapted according to the user's current interaction channel (such as APP pop-up, SMS, intelligent customer service dialog box, etc.) to ensure that the user can quickly obtain core information. At the same time, it supports the user to initiate subsequent operations based on the response information (such as confirming the plan, refusing to execute, consulting questions, etc.), forming a closed loop link of plan generation - response output - interactive feedback.

[0039] As described above, the resource transfer control method provided in this embodiment responds to service dialogue information input by the user terminal to the dialogue interaction agent, determines the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information; then, based on the target operation intention, it indexes the corresponding resource transfer service processing workflow from a preset job graph. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model agent with heterogeneous functions; subsequently, it drives at least one large model agent to sequentially execute each sub-task node in the resource transfer service processing workflow to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library; finally, based on the dialogue interaction agent, it generates resource transfer control response information according to the resource transfer control scheme and outputs it to the user terminal. In other words, on the one hand, by completing the closed loop of demand collection and result feedback through conversational intelligent agents, the resource transfer service has been transformed from being driven by manual operation to being driven by intelligent dialogue. Users can obtain resource transfer control solutions that meet their needs simply through natural dialogue, without having to manually fill out cumbersome forms or navigate through multiple links. This greatly simplifies the user's workflow during resource transfer, reduces user interaction costs, and improves the flexibility and user experience of resource transfer operations. On the other hand, by standardizing the workflow of resource transfer services through a job graph, and combining it with functionally heterogeneous large-scale intelligent agents to complete the specialized division of labor for sub-task nodes, this collaborative model not only achieves modular and standardized decomposition of the resource transfer service process, ensuring the standardization of process execution, but also effectively guarantees the processing accuracy and efficiency of each sub-task by relying on the heterogeneous capabilities of large-scale intelligent agents. This makes the execution logic of the resource transfer service more flexible and adaptable to different scenarios, providing reliable technical support for resource transfer services under different demand scenarios.

[0040] Based on the methods described in the above embodiments, please refer to Figure 4 , Figure 4 This is a flowchart illustrating another resource transfer control method provided for an exemplary embodiment of this application. The resource transfer control method includes the following steps S302-S314, wherein: S302. In resource transfer scenarios, obtain service dialogue information input by the user terminal for the dialogue interaction agent.

[0041] Please refer to the above for further details. Figure 3 Users can engage in at least one round of dialogue with the intelligent dialogue agent deployed on the service platform to input their specific needs. Interaction methods include voice, text, and graphical interfaces. The service platform can deploy a memory module to store historical interaction data between the intelligent dialogue agent and each user, including but not limited to historical dialogue data, historical decision data, and historical resource transfer control schemes (historical schemes). Historical dialogue data includes the context data of each complete dialogue between the agent and the user, fully preserving all content such as the expression of needs, questions, information confirmation, and scheme communication. Historical decision data may include various algorithmic, model invocation, model inference, and rule matching operations performed by the agent in response to each user's specific input needs, as well as core information such as the corresponding decision reasons, decision basis, and decision deduction process.

[0042] S304. Extract intent features from service dialogue information, calculate the confidence level of the intent features in a preset intent space, and determine the target operation intent based on the confidence level. The operation intents corresponding to the preset intent space include resource transfer intent, execution confirmation intent, and strategy optimization intent.

[0043] The dialogue interaction agent, leveraging its powerful semantic understanding capabilities, can extract key information related to resource transfer (such as resource transfer amount and resource return cycle) from service dialogue information and transform it into a structured intent feature vector, thereby completing the extraction of intent features. The preset intent space is a standardized intent classification space constructed based on various types of operations in the resource transfer scenario, including multiple typical operational intents in the resource transfer scenario, such as resource transfer intent, execution confirmation intent, and strategy optimization intent.

[0044] Confidence refers to the degree of matching between the extracted intent features and typical intent types in the preset intent space. It can be obtained by calculating the similarity between the extracted intent features and the feature template vectors of each typical intent type in the preset intent space, and then normalizing the obtained similarity values ​​to obtain the confidence of the intent feature corresponding to a certain typical intent.

[0045] Specifically, the target operation intent can be determined according to the confidence value distribution of each typical operation intent in the preset intent space and the preset intent determination rules. For example, if the confidence value of a typical operation intent is the highest among all intents, and the value is greater than or equal to the preset confidence threshold, then the typical operation intent can be determined as the target operation intent. If the confidence values ​​of all typical operation intents are lower than the preset confidence threshold, then the secondary interactive inquiry process of the dialogue interaction agent is triggered to collect additional dialogue information from the user to re-extract the intent features and calculate the confidence. If multiple typical operation intents have the same confidence value and are all the highest, the above-mentioned secondary interactive inquiry process is also triggered to confirm the user's real needs and complete the accurate determination of the target operation intent.

[0046] S306. Extract user explicit demand information from service dialogue information, determine the explicit demand type that matches the user explicit demand information from a preset demand type set, and use the remaining demand types as reference demand types. This demand type set includes resource transfer quota demand, user rights demand, resource transfer cost demand, user preference demand, and resource return demand.

[0047] Among them, the user-displayed demand information is the resource transfer-related demand information that is actively and explicitly expressed by the user from the full context of the service dialogue information. This type of information is the core demand content directly disclosed by the user.

[0048] Specifically, the extracted explicit user requirements information can be matched one by one with the core features of each type in the requirement type set to determine the requirement types that are completely or highly matched (the feature matching threshold reaches a certain value). These are classified as explicit requirement types that match the explicit user requirements information, while the rest are classified as reference requirement types.

[0049] For example, if a user's target operation intent during the dialogue is resource transfer, and the core demand corresponds to the matching demand type of resource transfer amount demand, then the resource transfer amount demand is the display demand type, and the remaining demand types, such as user rights demand, user preference demand, resource transfer cost demand, and resource return demand, are used as reference demand types.

[0050] S308. Based on the user-displayed requirement information, determine the reference requirement information corresponding to each reference requirement type, and combine the user-displayed requirement information and the reference requirement information to obtain at least one user requirement constraint combination.

[0051] Specifically, based on the core user demands represented by the explicit user demand information, and combined with the association rules of resource transfer operation scenarios and user demand matching logic, reference demand information adapted to the core user demands is derived and generated for each reference demand type, thus completing the full-dimensional completion of demand information; then, the explicit user demand information and the completed reference demand information are combined to form complete constraints containing the core user demands and reference demands.

[0052] Each user requirement constraint combination must include user-displayed requirement information, as well as at least one other type of requirement information. For example, user-displayed requirement information can be combined with various reference requirement information to obtain multiple user requirement constraint combinations. Each combination is based on user-displayed requirement information and is paired with reference requirement information of a single reference requirement type, forming a basic constraint combination with a concise structure and clear dimensions.

[0053] In some embodiments, step S308 above, "determining the reference requirement information corresponding to each reference requirement type based on the user-displayed requirement information," specifically includes: When the reference requirement type is a user preference requirement, the reference requirement information corresponding to the user preference requirement is determined according to the current user's historical resource transfer control scheme. When the reference requirement type is a requirement type other than the user's preferred requirement, the optional requirement range of each reference requirement type that matches the user's display requirement information is determined based on the preset requirement constraint rules, and the upper limit of the optional requirement range is used as the reference requirement information of the corresponding reference requirement type.

[0054] The optional demand range is a valid value range that aligns with the core user needs of this resource transfer service and complies with the basic rules of the resource transfer scenario. The upper limit of the optional demand range refers to the value within the valid value range that is practically feasible and maximizes user benefits. This value will be differentiated based on the attribute characteristics of the demand type. For example, for resource transfer amount, the upper limit is the highest supported transfer amount; for resource transfer cost, the upper limit is the lowest transfer cost; and for resource return cycle, the upper limit is the longest allowed return cycle.

[0055] The preset requirement constraint rules can be formulated based on the attribute information of each resource transfer option in the resource transfer option library. For example, when the resource usage scenario is scenario A, resource transfer options that are compatible with scenario A are extracted from the resource transfer option library, and the attribute values ​​of each option are specified. Then, the attribute values ​​are aggregated, filtered, and delimited to define the adaptation boundaries of each attribute value in scenario A.

[0056] S310. Based on the target operation intention, index the corresponding resource transfer service processing workflow from the preset operation map. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions.

[0057] The steps S310 and S204 are the same as those described above, and will not be repeated here.

[0058] S312. Drive at least one large model agent to sequentially execute each subtask node in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user requirement constraints and the resource transfer option library.

[0059] The steps S312 and S206 are the same as those described above, and will not be repeated here.

[0060] S314. Based on the dialogue interaction intelligent agent, generate resource transfer control response information according to the resource transfer control scheme, and output the resource transfer control response information to the user terminal.

[0061] For example, if a user's core demand during a conversation with the dialogue agent "Xiao X" is determined to be a target resource transfer amount, corresponding to a resource transfer amount demand, then if each demand constraint combination contains only two types of demand information, several demand constraint combinations can be formed: target resource transfer amount + maximum user rights, target resource transfer amount + user preferences, target resource transfer amount + minimum resource return requirements, and target resource transfer amount + minimum resource transfer cost. Then, at least one large model agent is driven to generate a corresponding resource transfer control scheme based on each demand constraint combination. Following the response template and script specifications corresponding to the resource transfer intent, resource transfer control response information embedded with guiding information is generated and displayed to the user in card format. This achieves the goal of recommending suitable solutions to the user from multiple dimensions based on their core demand.

[0062] As can be seen from the above, the resource transfer control method provided in this application takes the user-submitted demand information as the core, supplements it with other dimensions of compatible reference demand information, and combines the user demand information with the reference demand information to construct different constraints from multiple dimensions such as transfer amount, transfer cost, user rights, user preferences, and resource return requirements. In this way, while meeting the user's core demands, it can also take into account additional demands from different dimensions to accurately formulate an adaptation plan, realize the parallel output of multiple plans to provide users with a diverse selection space, and effectively improve the flexibility and scenario adaptability of resource transfer plan generation.

[0063] Based on the methods described in the above embodiments, please refer to Figure 5 , Figure 5 This is a flowchart illustrating another resource transfer control method provided for an exemplary embodiment of this application. The resource transfer control method includes the following steps S402-S414, wherein: S402. In a resource transfer scenario, in response to the service dialogue information input by the user terminal to the dialogue interaction agent, determine the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information.

[0064] The steps S402 and S202 are the same as those described above, and will not be repeated here.

[0065] S404. Based on the target operation intention, index the corresponding resource transfer service processing workflow from the preset operation map. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions.

[0066] If the target operation intent is a resource transfer intent or a solution optimization intent, the indexed workflow will involve core sub-task nodes such as filtering options within the library, parameter configuration, and solution generation. Solution optimization intent will also involve sub-task nodes such as parsing the original solution and matching optimization items. If the target operation intent is an execution confirmation intent, the indexed workflow will be the execution verification and implementation process of the resource transfer control solution, involving core sub-task nodes such as solution compliance verification, execution permission verification, transfer parameter review, execution instruction generation, and execution result feedback. Different sub-task nodes can be executed by a large model agent.

[0067] S406. Select the target resource transfer control item corresponding to each user requirement constraint combination from the preset resource transfer control item library, and generate the transfer control parameters of the target resource transfer control item.

[0068] The resource transfer control item library includes standardized resource products adapted to different resource transfer scenarios. Each product in the library is configured with basic attribute information, such as the resource transfer limit, transfer time period, resource transfer cost, resource return period, product rights, and risk level that the product can provide. This attribute information not only represents the product's access requirements but also clarifies the boundaries of the rights that the product can provide. Subsequently, the product can be accurately adapted based on the specific constraints of the user's needs and the product attribute information.

[0069] Transfer control parameters are executable, refined, and personalized parameter values ​​determined for the selected target resource transfer control item (standardized resource product). They are the core execution basis for the target resource transfer control item to be implemented as a resource transfer control scheme. These parameters do not exceed the parameter value range defined by the product attribute information, and can accurately match the specific constraints in the user's demand constraint combination, while meeting the platform's risk control threshold, regulatory compliance standards, and resource supply capacity requirements.

[0070] In some embodiments, step S406 specifically includes steps (1) and (2), wherein: (1) The driver library selects the intelligent agent execution library to select the sub-task node, and uses the first target large language model to perform multi-dimensional feature analysis on the user requirement constraint combination to obtain the feature analysis result, and matches the target resource transfer control item that is compatible with the feature analysis result from the preset resource transfer control item library.

[0071] In some embodiments, step (1) above specifically includes: The user requirement constraints are input into the first target large language model to construct a differentiated optimization objective function for this resource transfer service. Based on this differentiated optimization objective function, the corresponding target resource transfer control item is indexed in the preset resource transfer control item library.

[0072] The core function of the differentiated optimization objective function is to transform users' unstructured, multi-dimensional needs and constraints into a quantifiable and calculable optimization index system, providing a precise index for matching subsequent resource transfer control items. When constructing the differentiated optimization objective function, the First Objective Big Language Model maps the combination of user needs constraints with the basic attribute information of each option in the resource transfer option library. It first analyzes the user needs from multiple dimensions, including scenario features, constraint strength, priority weight, cost sensitivity, and risk tolerance, generating structured feature analysis results. Based on these results, it determines the core optimization indicators of the objective function (such as "minimizing transfer costs," "maximizing transmission efficiency," and "prioritizing the fulfillment of rights and interests"), and transforms the feature parameters of each dimension into variables, coefficients, and constraint boundaries of the objective function, forming a unique differentiated optimization objective function suitable for this resource transfer service. Each differentiated optimization objective function is configured with its own exclusive index matching rules and threshold judgment criteria. The index matching rules correspond to the attribute label system of each control item in the resource transfer control item library. The control items in the library have been pre-classified and archived according to core dimensions such as quota level, cost (price) level, risk level, equity type, and timeliness priority. For example, the timeliness priority objective function will prioritize indexing control items marked with the "immediate transfer" attribute in the library, and its matching threshold focuses on the fit of the timeliness dimension (the preset threshold is not less than 90%). The lowest cost objective function prioritizes indexing control items with the "zero transfer cost" and "transfer cost discount" attributes, and the matching threshold focuses on the fit of the cost dimension (the preset threshold is not less than 85%).

[0073] During the indexing and matching phase, the search criteria are the range of optimal solutions, core constraint parameters, and target optimization direction of the differentiated optimization objective function. Similarity calculations are performed between these criteria and the attribute tags and parameter ranges of each control item in the resource transfer control item library. Control items with a matching degree higher than a preset threshold are selected as candidate resource transfer control items. If multiple candidate control items exist, they can be further ranked based on the quantitative calculation results of the objective function. The control item with the highest overall suitability is selected as the final target resource transfer control item, ensuring that the matching results accurately match user needs and service scenario characteristics. Alternatively, the cumulative number of selections of each target resource transfer control item by the user group associated with the current user (associated users) can be further statistically analyzed, and the target resource transfer control item with the highest cumulative selection count is selected to generate the resource transfer control scheme. Associated users can be users with the same application qualifications, historical preferences, and / or basic personal information as the current user. Basic personal information includes age, gender, and / or geographical location. By integrating historical selection behavior data of related users, the matching scope of resource transfer control items can be effectively narrowed, and the accuracy and adaptability of the final solution can be improved. At the same time, relying on the common patterns of group selection, the rationality and implementation of the resource transfer control solution can be guaranteed.

[0074] (2) The driving parameter processing agent executes the parameter configuration sub-task node. For each target resource transfer control item, it uses the second target large language model to determine the recommended boundary parameter conditions that satisfy the corresponding user demand constraint combination, and determines the transfer control parameters of the corresponding target resource transfer control item based on the recommended boundary parameter conditions.

[0075] In some embodiments, step (2) above specifically includes: Based on the characteristics of the target resource transfer control item, a parameter calculation logic chain containing constraint variables and target variables is generated using the target large language model. The parameter calculation logic chain is then executed to perform parameter calculations within the numerical range limited by the user requirement constraint combination, thereby obtaining the transfer control parameters used to fill the target resource transfer control item.

[0076] Among them, the resource transfer control item characteristics refer to the core attributes inherent to each target control item and directly related to parameter configuration, including but not limited to the type of resource transfer channel supported by the control item, the upper / lower limit of the quota, the cost (price) billing benchmark, the timeliness requirement threshold, the risk prevention and control level, and the associated operation permission scope, etc. This characteristic will serve as the core basis for constructing the parameter calculation logic chain, ensuring that the logic chain adapts to the inherent attribute constraints of the control item. Before generating the parameter calculation logic chain, the second target large language model will first complete a two-dimensional association mapping: on the one hand, it will match the target resource transfer control item characteristics with the preset parameter template library and call the basic parameter calculation framework adapted to this type of control item; on the other hand, it will transform the demand information in the combination of user demand constraints (such as resource transfer quota demand, resource transfer cost demand, resource return demand, etc.) into constraint variables that can be substituted into the calculation, and at the same time set the transfer control parameters to be configured (such as specific transfer quota, transfer cost, return cycle, etc.) as target variables.

[0077] The parameter calculation logic chain comprises four core modules: variable definition, association rules, iterative calculation process, and verification standards. The variable definition module clarifies the value types, units, and relationships between constraint variables and target variables. The association rules module establishes quantitative association formulas between variables based on operational norms in resource transfer scenarios (such as transmission loss accounting rules in the power sector and bandwidth allocation rules in the computing power sector). For example, transmission loss = transmission power × transmission distance × loss coefficient. The iterative calculation process module performs multiple rounds of iterative calculations on the target variable within the numerical range limited by user-defined constraints, gradually narrowing the parameter value range to ensure that the calculation results satisfy both user constraints and adapt to control item characteristics. The verification standards module presets parameter validity thresholds (such as transfer costs not exceeding regulatory limits and transmission losses not exceeding industry standards), performing compliance and adaptability verification on the results of each iteration. If the results do not meet the verification requirements, the calculation coefficients are automatically adjusted and the iteration is repeated until a suitable parameter value is obtained.

[0078] The recommended boundary parameter conditions are the effective value ranges of each target variable output by the second objective large language model based on the above logical chain. This range takes user demand constraints as the core boundary, while also considering the characteristics of target resource transfer control items and scenario compliance requirements. When determining the final transfer control parameters, the parameter processing agent will combine the calculation results of the parameter boundary calculation engine to select the optimal value from the recommended boundary parameter conditions. That is, it will prioritize selecting the intermediate or optimal value that fits the user's core needs. If there are multi-dimensional parameter collaborative configuration requirements (such as the linkage adjustment of transfer amount and transfer cost, the balance of timeliness and risk control parameters, etc.), then collaborative optimization will be carried out through refined parameter adaptation rules to ensure that there are no conflicts between parameters and that the overall requirements of user needs and scenario implementation are met.

[0079] It should be noted that all of these large-scale intelligent agents can be trained on large language models, and some agents can also be trained on machine learning models. Different domains require different training data for model training. Optionally, the following is an example of a training process for the above-mentioned parameter processing agents applied to the target domain: First, a dedicated training dataset for resource transfer scenarios in the target domain is constructed. This dataset is based on industry regulatory standards, practical rules for various resource transfer operations, and historical parameter configuration cases in the target domain. It also includes labeled data related to user demand constraints and parameter configuration results, as well as error correction samples of parameter violations. The data is then graded, cleaned, and labeled to clarify the control item features, user demand constraint combinations, parameter calculation logic chains, recommendation boundary parameter conditions, and final transfer control parameters for each sample. At the same time, compliance verification points and parameter collaborative configuration rules are labeled to ensure the completeness, accuracy, and scenario adaptability of the dataset. Next, the second objective is to perform target domain pre-training adaptation of the large language model. Using a general large language model as the base model (including but not limited to DeepSeek large models, GPT series large models, etc.), the above-mentioned dedicated training dataset is transformed into a vector form that the model can understand. Through continuous pre-training, the model learns the professional terminology, parameter relationships, compliance boundary requirements and industry operation standards of the target domain resource transfer scenario. At the same time, quantitative calculation samples from the target domain are introduced to enhance the mathematical reasoning ability of the model, solve problems such as the construction of quantitative correlation formulas and the design of iterative calculation process in the generation of parameter calculation logic chain, and enable the model to have the ability to model parameter calculation logic in the target domain resource transfer scenario. Subsequently, model fine-tuning training was conducted, targeting three core sub-tasks: "parameter calculation logic chain generation, recommendation boundary parameter condition determination, and transfer control parameter selection and collaborative optimization." The target resource transfer control item features and user demand constraints were used as input prompts, and the adapted parameter calculation logic chain, compliant recommendation boundary parameter conditions, and appropriate transfer control parameter configurations were used as output labels. Supervised fine-tuning was employed to allow the agent to learn the mapping relationship between input and output. Simultaneously, few-shot learning and thought chain fine-tuning were introduced, embedding parameter calculation reasoning step annotations into the samples, enabling the agent to master the complete reasoning process of "feature matching - variable definition - rule construction - iterative calculation - compliance verification - parameter selection."

[0080] S408. Based on the combination of each resource transfer control item and the corresponding control item configuration parameters, generate at least one resource transfer control scheme.

[0081] Among them, the resource transfer control scheme is a systematic integration and scenario-based adaptation of each target resource transfer control item and the corresponding control item configuration parameters. Its core objective is to form a complete execution scheme that can be directly implemented, meets the full range of user needs, complies with scenario compliance requirements, and ensures that each control item works in coordination without conflict.

[0082] In this embodiment, the intelligent agent is selected by the driver library to filter sub-task nodes, the intelligent agent is executed by the driver parameter processing to configure sub-task nodes, and a resource transfer control scheme is generated based on the resource transfer control items and transfer control parameters. That is, the intelligent configuration of control items and parameters is completed by relying on the professional division of labor of the intelligent agents, thereby improving the accuracy and efficiency of control item selection and parameter generation, effectively avoiding the deviation and redundancy caused by manual configuration, and making the design of the resource transfer control scheme more professional, intelligent and practical.

[0083] S410. Based on the dialogue interaction agent, generate resource transfer control response information according to the resource transfer control scheme, and output the resource transfer control response information to the user terminal.

[0084] The steps S410 and S208 are the same as those described above, and will not be repeated here.

[0085] S412. If an adjustment feedback message is received for the resource transfer control response message, the adjustment feedback message is treated as new service dialogue information, and the resource transfer control scheme generated in the previous round is retained as context memory.

[0086] S414. Trigger the resource transfer service processing workflow corresponding to the execution strategy tuning intent to update the scheme parameters in the context memory based on the adjustment feedback information.

[0087] The target workflow corresponding to the strategy tuning intent is mainly used to adjust the parameters (such as the amount of data transferred and the cost of resource transfer) of existing resource transfer control schemes. For example, when a user has selected a resource transfer control scheme, the resource transfer service processing workflow corresponding to the strategy tuning intent can be triggered. Specifically, the adjustment feedback information can be semantically parsed and the requirements extracted first to clarify the type, target value, and adjustment range of the parameters to be adjusted, thus obtaining the user's requirement information. Then, the previous round of resource transfer control schemes stored in the context memory can be structurally parsed to locate the parameter nodes and associated configurations that match the optimization requirements. Subsequently, the target parameters are updated and configured according to the preset parameter adjustment rules, and the resource transfer option library is filtered based on the updated configuration parameters. Next, the input parameters for the scheme are generated based on the selected target resource options, and then a new scheme is generated. Finally, the completeness and compliance of the generated scheme are verified to ensure that the new scheme meets the operational rules and risk control requirements of resource transfer, and the new scheme data is synchronized to the context memory to complete the scheme iteration of this round of strategy tuning.

[0088] In this embodiment, this interactive scheme optimization mechanism can dynamically iterate and optimize the resource transfer control scheme based on real-time user feedback. At the same time, relying on the context memory mechanism, it effectively avoids repetitive operations and loss of key information during the scheme optimization process, greatly improving the execution efficiency and operational continuity of scheme optimization. This realizes the upgrade of resource transfer control from "one-time scheme generation" to "full-process interactive optimization", making the generated resource transfer control scheme more in line with the user's actual needs and scenario adaptability.

[0089] Based on the resource transfer control method described in the above embodiments, please refer to... Figure 6 This application also provides a resource transfer control device. The resource transfer control device 500 includes a determining unit 501, an indexing unit 502, a driving unit 503, and a reply unit 504, wherein: The determining unit 501 is used to, in a resource transfer scenario, respond to the service dialogue information input by the user terminal to the dialogue interaction agent, and determine the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information. The indexing unit 502 is used to index the corresponding resource transfer service processing workflow from the preset job graph based on the target operation intention. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions. The driving unit 503 is used to drive at least one large model agent to sequentially execute each sub-task node in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library. The response unit 504 is used to generate resource transfer control response information based on the dialogue interaction agent and the resource transfer control scheme, and output the resource transfer control response information to the user terminal.

[0090] In one possible implementation, the determining unit 501 is used for: Extract explicit user request information from the service's dialogue information; The set of explicit demand types is determined from the preset set of demand types to match the explicit demand information of the user, and the remaining demand types are used as reference demand types. The set of demand types includes resource transfer quota demand, user rights demand, resource transfer cost demand, user preference demand, and resource return demand. Based on the user's displayed requirements information, determine the corresponding reference requirements information for each reference requirement type; The user's displayed requirement information and the reference requirement information are combined to obtain at least one combination of user requirement constraints.

[0091] In one possible implementation, the determining unit 501 is used for: When the reference requirement type is a user preference requirement, the reference requirement information corresponding to the user preference requirement is determined according to the current user's historical resource transfer control scheme. When the reference requirement type is a requirement type other than the user's preferred requirement, based on the preset requirement constraint rules, the optional requirement range of each reference requirement type that matches the user's display requirement information is determined, and the upper limit of the optional requirement range is used as the reference requirement information of the corresponding reference requirement type.

[0092] In one possible implementation, the drive unit 503 is used for: Select the target resource transfer control item corresponding to each user requirement constraint combination from the preset resource transfer control item library, and generate the transfer control parameters of the target resource transfer control item. Based on the resource transfer control item and the corresponding transfer control parameters, a corresponding resource transfer control scheme is generated.

[0093] In one possible implementation, the drive unit 503 is used for: The driver library filters the intelligent agent execution library to filter the sub-task nodes, uses the first target large language model to perform multi-dimensional feature analysis on the combination of user requirements constraints to obtain feature analysis results, and matches the target resource transfer control items that are suitable for the feature analysis results from the preset resource transfer control item library; The driving parameter processing agent executes the parameter configuration subtask node. For each target resource transfer control item, it uses the second target large language model to determine the recommended boundary parameter conditions that satisfy the corresponding user demand constraint combination, and determines the transfer control parameters of the corresponding target resource transfer control item based on the recommended boundary parameter conditions.

[0094] In one possible implementation, the drive unit 503 is used for: The user requirement constraints are combined and input into the first target large language model to construct a differentiated optimization objective function for this resource transfer service. Based on the differentiated optimization objective function, the corresponding target resource transfer control item is indexed in the preset resource transfer control item library. Based on the characteristics of the target resource transfer control item, a parameter calculation logic chain containing constraint variables and target variables is generated using the target large language model. The parameter calculation logic chain is then executed to perform parameter calculations within the numerical range limited by the user requirement constraint combination, thereby obtaining the transfer control parameters used to fill the target resource transfer control item.

[0095] In one possible implementation, the determining unit 501 is used for: The intent features are extracted from the service dialogue information, and the confidence level of the intent features in the preset intent space is calculated. The operation intents corresponding to the preset intent space include resource transfer intent, execution confirmation intent, and strategy optimization intent. The target's operational intent is determined based on this confidence level.

[0096] In one possible implementation, the response unit 504 is also used for: After outputting the resource transfer control response information, if adjustment feedback information is received in response to the resource transfer control response information, the adjustment feedback information is used as new service dialogue information, and the resource transfer control scheme generated in the previous round is retained as context memory. Trigger the resource transfer service processing workflow corresponding to the execution strategy tuning intent, and update the scheme parameters in the context memory based on the adjustment feedback information.

[0097] As described above, the resource transfer control device 500 provided in this application embodiment, through the determining unit 501 responding to the service dialogue information input by the user terminal for the dialogue interaction agent, determines the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information; the indexing unit 502 indexes the corresponding resource transfer service processing workflow from a preset job graph based on the target operation intention, the resource transfer service processing workflow including multiple sub-task nodes arranged according to a preset logical topology and executed by at least one functionally heterogeneous large model agent; subsequently, the driving unit 503 drives at least one large model agent to sequentially execute each sub-task node in the resource transfer service processing workflow, so as to obtain at least one resource transfer control scheme by generating a scheme based on the combination of user demand constraints and the resource transfer option library; finally, the reply unit 504 generates resource transfer control reply information based on the dialogue interaction agent and the resource transfer control scheme, and outputs it to the user terminal. In other words, on the one hand, by completing the closed loop of demand collection and result feedback through conversational intelligent agents, the resource transfer service has been transformed from "manual operation-driven" to "intelligent dialogue-driven." Users can obtain resource transfer control solutions tailored to their needs simply through natural dialogue, eliminating the need to manually fill out cumbersome forms and navigate through multiple links. This greatly simplifies the user's workflow during resource transfer, reduces user interaction costs, and enhances the flexibility and user experience of resource transfer operations. On the other hand, by standardizing the workflow of resource transfer services through a job graph, and combining it with functionally heterogeneous large-scale intelligent agents to complete the specialized division of labor for sub-task nodes, this collaborative model not only achieves modular and standardized decomposition of the resource transfer service process, ensuring the standardization of process execution, but also effectively guarantees the processing accuracy and efficiency of each sub-task by relying on the heterogeneous capabilities of large-scale intelligent agents. This makes the execution logic of the resource transfer service more flexible and adaptable to different scenarios, providing reliable technical support for resource transfer services under different demand scenarios.

[0098] It should be noted that the division of modules in the above-described resource transfer control device is for illustrative purposes only. In other embodiments, the resource transfer control device can be divided into different modules as needed to complete all or part of the functions of the resource transfer control device. The implementation of each module in the resource transfer control device provided in this application embodiment can be in the form of a computer program. This computer program can run on a terminal or server. The program modules constituted by this computer program can be stored in the memory of the terminal or server. When the computer program is executed by a processor, it implements all or part of the steps of the resource transfer control method described in this application embodiment.

[0099] This application also provides an electronic device, which can be a server, and its internal structure diagram can be as follows: Figure 7As shown, this electronic device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. The processor executes computer programs to implement a resource transfer control method.

[0100] Those skilled in the art will understand that Figure 7 The structures shown are merely block diagrams of some structures related to the embodiments of this application and do not constitute a limitation on the electronic devices to which the embodiments of this application are applied. Specific electronic devices may include more or fewer components than those shown in the figures, or combine certain components, or have different component arrangements.

[0101] In one possible implementation, a computer storage medium is provided that stores instructions, which, when executed on a computer or processor, cause the computer or processor to perform one or more steps in the above embodiments. If the constituent modules of the above-described electronic device are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium.

[0102] In one possible implementation, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0103] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer storage medium or transmitted through the computer storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0104] It should be noted that the information (including but not limited to UI screenshots, annotation information, requirement description information, image data, and visual description files), data (including but not limited to data used for analysis, stored data, and displayed data), and signals involved in the embodiments of this application are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, the modification record information involved in this specification was obtained under full authorization.

[0105] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks. Unless otherwise specified, the technical features of this embodiment and its implementation can be combined arbitrarily.

[0106] The above-described embodiments are merely preferred embodiments of the present application and are not intended to limit the scope of the present application. Various modifications and improvements made by those skilled in the art to the technical solutions of the present application without departing from the design spirit of the present application should fall within the protection scope defined by the claims.

[0107] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A resource transfer control method, characterized in that, include: In the resource transfer scenario, in response to the service dialogue information input by the user terminal to the dialogue interaction agent, the target operation intention for the resource transfer service and at least one combination of user demand constraints are determined based on the service dialogue information. Based on the target operation intent, the corresponding resource transfer service processing workflow is indexed from the preset operation graph. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions. Drive at least one large model agent to sequentially execute each of the sub-task nodes in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library; Based on the dialogue interaction agent, resource transfer control response information is generated according to the resource transfer control scheme, and the resource transfer control response information is output to the user terminal.

2. The method as described in claim 1, characterized in that, Determining the target operation intent for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information includes: Extract explicit user request information from the service dialogue information; The set of demand types determines the explicit demand type that matches the user's explicit demand information from the preset demand type set, and uses the remaining demand types as reference demand types. The demand type set includes resource transfer quota demand, user rights demand, resource transfer cost demand, user preference demand, and resource return demand. Based on the user-displayed requirement information, determine the reference requirement information corresponding to each of the reference requirement types; The user display requirement information and the reference requirement information are combined to obtain at least one user requirement constraint combination.

3. The method as described in claim 2, characterized in that, The step of determining the reference requirement information corresponding to each of the reference requirement types based on the user display requirement information includes: When the reference demand type is a user preference demand, the reference demand information corresponding to the user preference demand is determined according to the current user's historical resource transfer control scheme. When the reference requirement type is a requirement type other than the user preference requirement, based on the preset requirement constraint rules, the optional requirement range of each of the reference requirement types that is compatible with the user display requirement information is determined, and the upper limit of the optional requirement range is used as the reference requirement information of the corresponding reference requirement type.

4. The method as described in claim 1, characterized in that, The process of generating at least one resource transfer control scheme based on the user requirement constraint combination and the resource transfer option library includes: Select the target resource transfer control item corresponding to each user requirement constraint combination from the preset resource transfer control item library, and generate the transfer control parameters of the target resource transfer control item; Based on the resource transfer control items and the corresponding transfer control parameters, a corresponding resource transfer control scheme is generated.

5. The method as described in claim 4, characterized in that, The step of selecting target resource transfer control items corresponding to each user requirement constraint combination from a preset resource transfer control item library and generating transfer control parameters for the target resource transfer control items includes: The driver library filters the intelligent agent execution library to filter sub-task nodes, uses the first target large language model to perform multi-dimensional feature analysis on the combination of user demand constraints to obtain feature analysis results, and matches the target resource transfer control items that are compatible with the feature analysis results from the preset resource transfer control item library. The driving parameter processing agent executes the parameter configuration subtask node. For each of the target resource transfer control items, it uses the second target large language model to determine the recommended boundary parameter conditions that satisfy the corresponding user demand constraint combination, and determines the transfer control parameters of the corresponding target resource transfer control item based on the recommended boundary parameter conditions.

6. The method as described in claim 5, characterized in that, The step of using a first target large language model to perform multi-dimensional feature analysis on the combination of user demand constraints to obtain feature analysis results, and matching target resource transfer control items that match the feature analysis results from a preset resource transfer control item library, includes: The user requirement constraints are combined and input into the first target large language model to construct a differentiated optimization objective function for this resource transfer service. Based on the differentiated optimization objective function, the corresponding target resource transfer control item is indexed in the preset resource transfer control item library. For each of the target resource transfer control items, the second target large language model is used to determine the recommended boundary parameter conditions that satisfy the corresponding user demand constraint combination, and the transfer control parameters of the corresponding target resource transfer control items are determined based on the recommended boundary parameter conditions, including: Based on the resource transfer control item characteristics of the target resource transfer control item, a parameter calculation logic chain containing constraint variables and target variables is generated using the target large language model. The parameter calculation logic chain is then executed to perform parameter calculations within the numerical range defined by the user requirement constraint combination, thereby obtaining transfer control parameters used to fill the target resource transfer control item.

7. The method according to any one of claims 1-6, characterized in that, Determining the target operation intent for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information includes: Intent features are extracted from the service dialogue information, and the confidence level of the intent features in a preset intent space is calculated. The operation intents corresponding to the preset intent space include resource transfer intents, execution confirmation intents, and strategy tuning intents. The target operational intent is determined based on the confidence level.

8. The method according to any one of claims 1-6, characterized in that, After outputting the resource transfer control response information, the resource transfer control method further includes: If an adjustment feedback message is received in response to the resource transfer control reply message, the adjustment feedback message is used as new service dialogue information, and the resource transfer control scheme generated in the previous round is retained as context memory. Trigger the resource transfer service processing workflow corresponding to the execution strategy optimization intent to update the scheme parameters in the context memory based on the adjustment feedback information.

9. A resource transfer control device, characterized in that, include: The determining unit is used, in a resource transfer scenario, to respond to service dialogue information input by the user terminal to the dialogue interaction agent, and to determine the target operation intention for the resource transfer service and at least one combination of user demand constraints based on the service dialogue information. An indexing unit is used to index the corresponding resource transfer service processing workflow from a preset job graph based on the target operation intention. The resource transfer service processing workflow includes multiple sub-task nodes arranged according to a preset logical topology and executed by at least one large model intelligent agent with heterogeneous functions. A driving unit is used to drive at least one large model agent to sequentially execute each of the sub-task nodes in the resource transfer service processing workflow, so as to generate at least one resource transfer control scheme based on the combination of user demand constraints and the resource transfer option library. The response unit is used to generate resource transfer control response information based on the dialogue interaction agent and the resource transfer control scheme, and to output the resource transfer control response information to the user terminal.

10. An electronic device, characterized in that, include: Processor and memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code stored in the memory to perform the method as described in any one of claims 1-8.

11. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions adapted for loading by a processor and executing the method as described in any one of claims 1-8.

12. A computer program product containing instructions, characterized in that, When the computer program product is run on a computer or processor, it causes the computer or processor to perform the method as described in any one of claims 1-8.

Images